CJI RESEARCH • AI INNOVATION
‘AI is going to be everywhere’
Artificial intelligence promises the next societal platform shift. But where does AI fit in business aviation? Words: Yves Le Marquand
CJI RESEARCH • AI INNOVATION
‘AI is going to be everywhere’
Artificial intelligence promises the next societal platform shift. But where does AI fit in business aviation? Words: Yves Le Marquand
“AI IS GOING to have an impact on the world at a scale no one understands.” That was how former Google CEO Eric Schmidt summarised the impact of AI a few months ago at Stanford University. With enough “spirit of full disclosure” to give a PR exec grey hair, Schmidt’s candid interview was swiftly deleted from Stanford’s YouTube channel. All mention was also initially swept from Google too. However, a copy does remain on, ironically, the Google-owned video streaming platform to which it was originally uploaded.
Schmidt believes the effects of AI on society will be much bigger than what he terms as the “horrific” impact of social media. Describing how large language models (LLMs), such as ChatGPT or Claude, learn today, Schmidt says systems remember like humans do.
“You ask it a question: read 20 books. You give it the text of the books and say tell me what they say. It forgets the middle, which is exactly how human brains work too. That is where we are.
“But there are people now building essentially LLM agents. They will tell an LLM to read something like chemistry, allow it to discover the principles and then test it, before adding that [learning] back into the LLM’s understanding. That is extremely powerful.”
The term AI is already everywhere. The browser even pulled up an advert for AI-powered web design when loading up the Schmidt interview. In a post-Covid world where promising startups across a diverse set of industries with a legacy of funding are struggling to fill rounds, AI has been the outlier. AI companies have been raising capital at such a rate over the past year that many haven’t made the effort to hire investment banks, even for multi-billion-dollar transactions.
On a recent podcast, Doug Leone, former managing partner at Sequoia Capital, said he too thinks AI is the next “platform shift” – in much the same way the mobile or the internet transformed society. “But we are going to overestimate it in the short term,” he cautions. “You see people are investing in everything. It was the same in ’99, but then Google came out of that or Facebook. There is no room to practice FOMO in this space.”
Usually on the conservative end of the technology adoption curve, business aviation as an industry typically follows Leone’s advice. When CJIQ charted the growth of digitisation in the industry last year, it found fewer than 200 companies had adopted digital processes in the front or back office.
But time (especially in terms of AI development) moves fast. Fast forward a year and it is possible to find companies including brokers, safety manuals, management platforms, operators and booking platforms that have all integrated some form of AI into their operations.
AI for optimisation
One of the pioneers of AI usage in the industry is Aerogility, a company that for the past 15 years has developed model-based AI systems using customer-specific datasets to help manage aircraft or engine availability and plan for all possible futures. Designed to be scalable to suit the size of operation, Aerogility boasts customers including Rolls-Royce and Lockheed Martin.
“It would be entirely implausible for AI to not be of value to the business aviation industry. You might say if people are not looking at it, then they are being deliberately inefficient because why would you not?” Simon Miles, head of AI at Aerogility tells us.
“Let’s imagine you are a charter company. You are going to need to do maintenance for a range of different aircraft. You have to make use of MRO facilities of which there is limited capacity, different costs and different staffing requirements. There are many factors which lead into what you could do and when. The question is how do I optimise this? It is a set of factors that grows more complex as your assets increase.”
To do this Miles and his team build what is known as a digital twin of the operational aspects of a client’s business. Using an airline as an example, Aerogility’s system will require information on the current fleet, time flown and state of the parts onboard. It will then layer data such as predicted aircraft movements, scheduling and average flight times to gauge forecasted wear and tear on the aircraft. Finally, the system requires data on the client’s maintenance capabilities, supply chain of replacement components and staff availability.
“With all of that we can create a mirror image in the virtual world of how your business is going to play out. We have customers such as Rolls-Royce and Lockheed Martin which provide us with very large sets of assets and we can predict events from a few months down the line to as far as 10 years away,” says Miles.
The Aerogility team use model-based AI systems rather than generative AI, which has been grabbing headline attention, because there is a significant increase in the reliability of the response.
Miles explains: “Model-based AI is more like deductive reasoning. You take what you know, such as how a particular MRO facility operates and the current state of a fleet and produce a strategic response. Even if something unexpected like a bird strike happens, the data will show the most effective response plan. There is a lot more reliability and straightforwardness in the path you take from start point to end point.
“For fleet optimisation, planning and strategic decision-making, this makes sense. But if you are looking at an alternative problem like detection of faults, then it becomes different because the user doesn’t know everything going on inside a component. You have a massive amount of sensor data and that is why something like machine-learning is appropriate to that particular problem.”
MySky first deployed AI in 2016
Another early adopter of AI systems in business aviation is MySky, a Switzerland-based company that developed a proprietary online audit platform for private jet operators using AI back in 2016. Fast-forward eight years and the firm has launched AI-based tools in multiple areas of its business from spend management and accounting operations to crew receipt submission.
“Our spend management platform is designed to do about 85-95% of the accounting tasks that people do on a daily basis at a flight operation,” Jean de Looz, head of Americas for MySky tells us. “This can be critical for Part 91 flight operations because that is an area of the industry which is typically tough financially and it really scales with the volume of flight operations.
“For example, it takes 10 to 12 minutes for someone to process an invoice or receipt. Most aircraft generate about 150 documents per month. If they’re in Part 135 it could be more, if they’re in Part 91 it could be less. If you do the math, it would take a human about 25 hours to process a month’s worth of documents. That gives you a scalability problem. Being able to remove that bottleneck using AI is super material to a business.”
In his time in business aviation de Looz has chartered, sold and flown aircraft. In his time in charter sales, it was commonplace to not know the profit and loss of a flight until at least 30 days after the trip (sometimes up to 45 days).
“We can create a mirror image in the virtual world of how your business is going to play out”.
Simon Miles, head of AI at Aerogility
“This is because you have to gather all the receipts and invoices which then have to be processed by accounting. I think a final P&L [profit and loss] on an international charter trip can take you up to 60 days on occasion, because it will continually evolve,” says de Looz. Using MySky Quotes, a proprietary AI-based charter management platform, a forward looking quote can be generated in a matter of seconds. “People made estimates before, but you probably had a 20% margin for error and had to 30 to 45 days for the final number. Now you have a four-second turnaround time and 5% margin for error.”
AI in aviation safety manuals
Swedish software company Web Manuals deployed an AI system it calls Amelia earlier this year to optimise its all-in-one document management system. Pioneering the use of AI for pilots’ manuals, Amelia allows users to pose questions and receive descriptive answers from within their manuals, helping them make “safe and well-informed” decisions while avoiding “costly” diversions and delays.
“When we first saw ChatGPT coming out my gut reaction was: ‘I don’t know if this is relevant for us’. Manuals are safety critical components and I wouldn’t necessarily trust an AI tool to write them for you,” Martin Lidgard, Web Manuals CEO tells CJI. “But AI is here to stay. Just like IT has become a natural component in daily life, AI will become a natural extension of IT and connectivity.”
Lidgard is also a firm believer that aviation is “already there” when it comes to the use of AI. Only its usage is becoming more apparent following rapid progress since the successful passing of the Turing test – a test of a machine’s ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human – in 2014 when Google’s chatbot Eugene Goostman convinced 33% of judges that it was human.
MySky's Jean de Looz says adopting AI, such as the systems deployed in MySky's spend management system, can perform about 85-95% of the daily accounting tasks usually assigned to staff in the back office.
“There are multiple use cases where AI analyses data to find different types of patterns to optimise complex technologies. This is already in use in business aviation, as well as in the development of more complex aircraft – autopilots etc. So, we are already there, but AI getting a voice and a face is why it has become so big in recent years,” says Lidgard.
“The cost of development is going down, thanks to AI itself which can code on your behalf.”
Ganesh Vedarattiname, director of Strategy & Analytics at Jetcraft
In the spirit of Volvo’s three-point seatbelt invention, Web Manuals has made its Amelia tool available to customers for an extra €1 per month per reader on top of their current subscription. “We see it as a tool that adds lots of extra value, but our focus is always on improving flight safety. Amelia is essentially something that at this stage enhances the user’s ability to find otherwise hard-to-find information in their manuals. I don’t think that should be something operators have to question whether or not they can afford. Safety improvements should be accessible to everyone, that is our philosophy here.”
A broker’s view
Ganesh Vedarattiname, director of Strategy & Analytics at Jetcraft has been encouraged by the hype surrounding generative AI because it has put the technology under the spotlight.
“As a result, the industry is trying to figure out how to deploy more advanced technology into its processes. This is vital to business aviation, which is, by nature, quite conservative. There’s an upward trend in technology investment which is ultimately going to make business aviation more efficient and able serve our clients better,” Vedarattiname tells CJI.
Jetcraft has been at the forefront of cutting-edge technology deployment for around a decade with highlights including a launch into the Metaverse in 2022 when it became the first business aviation company to purchase real estate in Decentraland – a 3D virtual world browser-based platform that allows users to buy virtual plots of land as non-fungible tokens (aka NFTs).
So far Jetcraft has integrated two AI systems into its daily operations. The first is a generative AI LLM built in-house called JetGPT. This ‘AI analyst’ collects information about the aircraft Jetcraft has for sale to allow the company to respond faster to a client’s request. The second utilises machine learning algorithms to look at market data, flight data and financial information to better understand fields and patterns.
Aerogility's team build what is known as a digital twin of the operational aspects of a specific business,“with all of that we can create a mirror image in the virtual world of how your business is going to play out," says Miles.
Aerogility's team build what is known as a digital twin of the operational aspects of a specific business,“with all of that we can create a mirror image in the virtual world of how your business is going to play out," says Miles.
A lack of structured data across much of business aviation can act as a deterrent to those looking to deploy AI systems, according to Vedarattiname. Without an organised dataset, AI all but loses the right to be branded intelligent. Also, the fragmented nature of often niche sub-sectors of business aviation means the costs of organising a dataset and building a bespoke system outweigh benefits for smaller operations.
“Due to the amount of fragmentation, what you see is AI technology concentrated amongst the top players. But I do think that is going to change in the short term for two reasons,” says Vedarattiname. “The cost of development is going down, thanks to AI itself which can code on your behalf now. Secondly, generative AI makes technology more accessible. You can talk to it as a human and it can explain really complex concepts in simple terms.”
But what are the benefits for the end user? Vedarattiname says first and foremost Jetcraft customers and employees note the increase in speed. “The pace at which we can process data, create a valuation and get market summaries has increased significantly. But on top of that we can also serve our clients better. The beauty of AI is it can analyse huge amounts of data on a daily basis. A human can just connect the dots. When you think about it, that is what the broker does trying to match a buyer and seller. AI can help do this on a massive scale.”
“...AI is here to stay. Just like IT has become a natural component in daily life, AI will become a natural extension of IT and connectivity.”
Martin Lidgard, CEO and founder, Web Manuals
Scale will change all
As Vedarattiname points out the continued increasing investment and scaling of AI systems will bring the barrier to entry down. In turn, allowing smaller operations to adopt AI into their operations. Key to assisting that transition are early adopters and pioneers – like Aerogility, MySky or Web Manuals – by keeping their subscription costs down and ensuring systems are scalable to operations of any size.
“Scalability is a big issue for businesses,” says MySky’s de Looz. “It is also for us. If we can make our system scalable, that allows our clients to scale themselves. And we allow them to become more competitive and adaptable at the same time,” he says. “Competitive is the perfect example. You can have a legacy management company that has spent millions of dollars developing internal reporting systems. We can overnight give MySky Quotes, which is turnkey and more capable at a fraction of the cost.”
De Looz also thinks AI is closing the gap between small and medium-sized operators and legacy giants. “If you are a legacy company that has put five million into your bespoke system, it is hard to back out of that investment and change a cardinal direction. This gives your small and medium-sized operators that chance to adopt a turnkey system today to compete at levels beyond legacy carriers.”
Axis is keen not to reinvent the wheel, it has developed its own in-house AI system, but cooperates with third party providers when it "makes sense", according to Mumenthaler.
AI in charter
Niall Olver, the entrepreneur responsible for the early success of ExecuJet, has jumped back into business aviation with a tech-based startup, AXIS Aviation. Launched in 2022, the aircraft management and charter firm offers data-driven asset management via a proprietary end-to-end data platform. Since the launch of its digital solution at EBACE this year, the firm has been working on incorporating AI into its user platform and will debut the latest update during NBAA-BACE 2024 in Las Vegas.
The company is taking a crawl, walk, run approach to AI integration that will eventually see it deploy AI-based systems throughout its operation, according to AXIS Aviation accountable manager, Kerstin Mumenthaler. Like Web Manuals’ Amelia tool, AXIS Aviation has aimed to make life easier for pilots and crews first.
“This means we have integrated all documentation, every manual – you can imagine how many pages that is – into an AI assistant for pilots and crews,” Mumenthaler tells us. “This allows crews to get the correct answer in seconds without needing to know which document or manual the information is contained within.”
Despite developing its own system in-house since 2022, the AXIS Aviation team cooperates with third-party providers when it makes sense. For example, the firm has a partnership with a Barcelona-based company that has enabled them to develop an “entirely new” AI-based system to streamline their operations. “Our idea is not to reinvent a new world, but to integrate the correct tools into our platform and build our own where necessary,” says Mumenthaler.
Despite AXIS Aviation’s progress into digitalisation and the deployment of AI systems, Mumenthaler thinks aviation as a whole is slow to adopt new technology. Jetcraft’s Vedarattiname noted the fragmentation in business aviation as a barrier to the wider deployment of AI. Mumenthaler believes this fragmentation is on purpose: “We want to do exactly the contrary. We want to be transparent and less fragmented. That is our business model, we take these [AI] tools which allow us to be more transparent.
“We are seeing it outside aviation already. If you talked a year ago about Chat GPT you received a lot of raised eyebrows. Nowadays everybody knows it, and most people are using it. Once aircraft owners begin to see the benefits, the mindset change will begin at their level and trickle down through the industry.”
Despite the clear advantages with respect to speed and efficiency to complete certain tasks, Mumenthaler understands why people might be apprehensive of AI. Reflecting ex-Google CEO Eric Schmidt’s top line, she says: “I think today we do not have any idea of what new possibilities will open up with support of AI. I think we have to be open minded. That said, I couldn’t decide on my own whether I found it scary or totally fascinating when I first used our AI assistant. It did in seconds what takes me half an hour. That is scary so I understand why people want to be careful with it, but on the other hand what we can do with it means I think there is no way not to integrate it in some way.”
“Our idea is not to reinvent a new world, but to integrate the correct tools into our platform and build our own where necessary.”
Kerstin Mumenthaler, accountable manager, AXIS Aviation
AI searches use up to 10x more power than a Google search. Data centres in the US will use 8% of domestic power by 2030, compared with 3% in 2022. Photo: Shutterstock
AI searches use up to 10x more power than a Google search. Data centres in the US will use 8% of domestic power by 2030, compared with 3% in 2022. Photo: Shutterstock
Negatives of AI
Distrust of AI, apprehension of technology or fear of the unknown are all potential drawbacks while society as a whole builds confidence in using AI systems. This article has largely discussed the positives of deploying of AI, but what are the negatives?
According to Aerogility’s Miles, transparency in decision-making is a drawback for AI systems generally. Knowing how an AI system got to its answer doesn’t matter so much if the prompt relates to a general knowledge question or the construction of a social media post. But that is not the same for a fleet management or maintenance scheduling operation for example.
“We pride ourselves in having a safe and trustworthy system that is not going to produce unjustifiable results. With Aerogility, you can always dig into the data to understand why it reached such a conclusion. But how to solve this in general is still an open question, given the large amount of reasoning an AI performs and the volume of data it produces,” says Miles.
“The difficulty with a machine learning system is you can't drill down because of the black box. This means you put data in and you get data out without having access to the reasoning in between. The best you can do is to try to adjust things and see whether you get a different outcome.”
There is also the potential for societal negatives as a result of AI, notes Web Manual’s Lidgard. He fears the trend of mass layoffs as companies create efficiencies through the deployment of AI systems will continue to accelerate. “As we introduce one technology, suddenly you can build the next one on top of it, and the next one and so on. You can make all sorts of philosophical conclusions from that, but not all those outcomes are necessarily better than before,” says Lidgard.
Society better get used to AI quickly because its integration into daily life is “inevitable”, according to MySky’s de Looz.
“AI is going to be everywhere. It is a platform change, similar to an industrial revolution,” he says. “We don’t really have a choice; it forms part of our eternal quest for greater and greater technology. Thinking about business aviation in particular, AI certainly has its place. But adoption rates will vary depending on what area of the industry we are talking about. Generative AI will eventually complete all tasks that are routine in nature from accounting to social media.”
Quantum mechanical effects, such as superposition, enable quantum computers to solve certain types of problems faster than traditional computers.
